Welcome

Thanks for your interest in our work!

Prerequisite

This instruction assumes that you are familiar with Java (JDK 1.7) and Eclipse IDE.

Maven users

If you are familiar with Maven environment, then you can just clone (or download) the repository and run mvn install to get vizweb-1.0-SNAPSHOT. See below on how to use the library.

Quick start

If you just need to run the compiled jar file and you are not interested in looking at the code, you can download the eclipse project here. Note that we use JDK 1.7 and we have not tested if it works with other Java versions.

Using the library

Before using the code, you should already have an image of a web page screenshot which will become the input.

There are three classes that provide methods for extracting the image features, namely ColorFeatureComputer, QuadtreeFeatureComputer and XYFeatureComputer. First, you need to import them.

#!java
import org.vizweb.ColorFeatureComputer;
import org.vizweb.XYFeatureComputer;
import org.vizweb.QuadtreeFeatureComputer;
import org.vizweb.quadtree.Quadtree;
import org.vizweb.structure.Block;

The following lines of code shows how to use them.

#!java
// the input is in the form of BufferedImage object
BufferedImage input = ImageIO.read(new File("images/awn.png"));

/*******************************
 * Compute color features
 *******************************/
double col = ColorFeatureComputer.computeColorfulness(input);
System.out.println("Colorfulness (method 1): " + col);

double col2 = ColorFeatureComputer.computeColorfulness2(input);
System.out.println("Colorfulness (method 2): " + col2);

/*******************************
 * Compute xy decomposition features (it may take a while)
 *******************************/
Block root = XYFeatureComputer.getXYBlockStructure(input);
System.out.println("Number of leaves: " + XYFeatureComputer.countNumberOfLeaves(root));
System.out.println("Number of group of text: " + XYFeatureComputer.countNumberOfTextGroup(root));

/*******************************
 * Compute quadtree features
 *******************************/
Quadtree qtColor = QuadtreeFeatureComputer.getQuadtreeColorEntropy(input);
System.out.println("Horizontal Balance: " + QuadtreeFeatureComputer.computeHorizontalBalance(qtColor));
System.out.println("Horizontal Symmetry: " + QuadtreeFeatureComputer.computeHorizontalSymmetry(qtColor));

Contact

Please send any feedback or question to mardiko at cs dot umd dot edu. Thanks!

Wiki

vizweb / Home